Cytochrome p450 metabolic site prediction method for small molecule drugs

Abstract

The present invention provides a method for predicting cytochrome P450 metabolic sites of small molecule drugs, using the machine learning model WhichCyp based on support vector machine classifiers, for small molecules belonging to cytochrome P450 enzyme subtypes 1A2, 2C9, 2C19, 2D6 and 3A4 Predict the substrates of one or several subtypes of the enzymes; use the convolutional neural network-based machine learning model to predict and rank the metabolic sites of small molecule drugs for the corresponding subtypes of cytochrome P450 enzymes; Calculate and evaluate the thermodynamic and dynamic interactions between the P450 enzyme system and the complete molecule; perform higher-precision MMGBSA calculations for each conformation to obtain the binding energy between different small molecule conformations and cytochrome P450 enzymes; use the collected small molecules The training set trains the process until the prediction accuracy is >80%. The invention improves the prediction accuracy.

Description

technical field [0001] The invention belongs to the technical field of drug metabolism, in particular to a method for predicting cytochrome P450 metabolic sites of small molecule drugs. Background technique [0002] WhichCyp is a support vector machine classifier-based machine learning model developed with PubChem Bioassay 1851 as the data set. The model can predict that small molecules belong to one or more substrates of cytochrome P450 enzyme subtypes 1A2, 2C9, 2C19, 2D6 and 3A4. [0003] SMARTCyp calculates the energy barriers between 139 small molecule fragments and the catalytic reaction centers of cytochrome P450 enzymes, forming a small database. When predicting the metabolic sites of small molecules, the energy barrier ranking of each site of small molecules is calculated according to the corresponding fragments. This method predicts cytochrome P450 enzyme 2D6, 2C isoforms, mainly based on the atoms and COO in small molecules - and NH 3 + The distance of the ene...

AI Technical Summary

Problems solved by technology

[0007] There are many approximations in the existing prediction methods, and when predicting the metabolic sites of small molecules by different cytochrome P450 enzyme subtypes, the prediction is only based on the structure of the small molecule, without considering the catalytic reaction centers of different subtypes of P450 enzymes. Influence of nearby residues on metabolic sites

Images